Abstract: As to the resilience of the supply chains, the importance of the timeliness of the supply recovery is no less than the recovery of the supply ability itself. This paper addresses the optimal design of the resilient distribution network oriented recovery time, in which a three-echelon distribution network is considered and DCs are backup DCs of each other. When a DC is broken down and cannot service its customer, the manufacturer maintains supplies through the nearest DC to the broken one. Thus DCs must hold some emergency inventory to meet customers’ demand for the arrival of emergent transshipments. A mixed integer linear programming model is constructed to formulate the problem. Because of the special structure of the model, the big M constraints, it takes much computational time to solve the model with optimization software, such as CPLEX, even with a high performance computing platform. To improve the efficiency of solving, a novel MILP decomposition method is proposed. With the experiments, the effectiveness and efficiency of the method are proved. At last, the sensitive analysis is made to understand the impact of various parameters on the distribution networks.

Key words: resilience, distribution network, timeliness, MILP decomposition method

摘要： 关于供应链的弹性,供应能力恢复时效性的重要程度丝毫不亚于供应能力的恢复程度。本文研究了面向供应恢复时效性的弹性分销网络设计问题。考虑一个三级分销网络,其中各分销中心互为各自的备用分销中心。当一个分销中心因发生失效事件而不能接受制造商的供货时,制造商通过其他分销中心向失效的分销中心供货。分销中心需要保持一定量的应急库存以满足在应急供应到达之前的顾客需求。针对该问题建立了混合整数规划模型。该模型包含的大M约束使得直接使用优化软件求解需要很长的计算时间,并对计算平台的硬件配置有较高要求。为了更高效地求解该模型,提出了改进的MILP分解算法。通过对比实验,证明了所提算法的效率和有效性。进行了模型的敏感性分析,研究了各项参数对分销网络结构和成本的影响。

关键词: 弹性, 分销网络, 时效性, MILP分解算法